Unbalanced Atrioventricular Septal Defect – A Congenital Heart Surgeons’ Society Inception Cohort Study

Table of Contents

  1. Abstract
  2. Specific Aims

a.Objectives

b.Hypothesis

  1. Background and Rationale
  2. Study Design
  1. Enrollment
  2. Echo Core Lab
  1. Participation Criteria
  1. Echocardiography Training
  2. Member Participation
  1. Study Population
  1. Inclusion Criteria
  2. Exclusion Criteria
  3. Number of subjects projected
  1. Study Endpoints and Evaluation
  1. Primary Endpoints
  2. Secondary Endpoints
  3. Statistical methods
  4. Demographic Data Points
  5. Clinical Documents and Studies
  6. Clinical Data Points
  1. Appendices
  1. Appendix 1: Definition
  1. References
  1. Abstract

Context: Atrioventricular Septal Defect (AVSD) is a rare congenital cardiac malformation of the atrioventricular septum. It results in a common atrio-ventricular valve orifice that connects both atria to both ventricles. Typically the connection is symmetrical thereby allowing equal balanced blood flow into each ventricle. But there is a spectrum of unequal atrio-ventricular connection associated with underdevelopment of either left or right ventricle. These neonates have an unbalanced flow and that in the extreme situation results in a heart that is functionally a single ventricle. The degree of unbalance affects both the type and risk of operative repair. The criteria that define the limits of operative repair and risk are unknown.

In our preliminary retrospective multi-center analysis we identified criteria that define the amount of unbalance. Our proposed prospective study of a multi-center inception cohort of neonates born with complete AVSD will define the limits of unbalanced flow and ventricle development to determine the optimal surgical treatment and outcomes.

Hypothesis: Survival, morbidity and functional outcomes of newborns with AVSD can be optimized through improved matching of repair strategy to morphologic/physiologic substrate.

Study Design: We will enroll a prospective cohort of infants with complete AVSD at participating CHSS member institutions. The CHSS Data Center at The Hospital for Sick Children in Toronto will collect and abstract clinical data and obtain copies of initial cardiac Echocardiograms & CT/MRI studies for blinded review. An annual cross-sectional follow-up of the cohort, including details of future tests and intervention will be entered into the dataset.

Study Measures: Longitudinal multivariate analyses of demographic, cardiac morphology, function, and procedural variables will be used to search for risk factors that affect outcomes. These will include analyses of recurring events such as follow-up echocardiographic/CT/MRI measures of heart function and re-interventions.

2. Specific Aims

2.a Objective: To improve survival among patients with atrioventricular septal defects (AVSD) by further characterizing that portion of the disease spectrum customarily referred to as unbalanced atrioventricular septal defect (uAVSD) and evaluating the relationships between patient and procedural factors and outcomes.

The AVSD is a spectrum of disease thatincludes subcategories characterized byvarying degrees of malalignment of the common atrioventricular (AV) junction, ventricular hypoplasia, and intrinsic valvar abnormalities. These morphologic features,in combination or in isolation, may result in disproportionate flow into right and left ventricles. This condition is commonly referred to as “unbalanced” atrioventricular septal defect.

Proper selection of treatment strategies for uAVSD is particularly difficult. Consensus regarding a standard definition of “unbalance” is lacking, and there are few evidence based guidelines for selection of treatment strategies. The overall high degree of mortality observed in patients with uAVSD is likely to reflect suboptimal choices of treatment strategy. Thus, overall survival might be better if the relationships between morphologic and physiologic aspects of the disease and the outcomes of various surgical treatment strategies were more fully understood. Treatment choices are relatively clear at the extreme ends of the anatomic spectrum of disease. When the AVSD is severely unbalanced, the smaller ventricle is incapable of supporting adequate cardiac output and functionally univentricular repair is required. In contrast, a biventricular repair strategy is uniformly appropriate for balanced AVSD. Between these extremes, the choice of repair strategy is confounded by gaps in present knowledge of the relationship between patient factors, anatomy, repair strategy, and outcome.

2.b Hypothesis: Survival and late outcomes of patients with uAVSD can be optimized through improved matching of repair strategy to morphologic/physiologic substrate.

Aim 1 (A1): Define the anatomic features of uAVSD.

SubAim 1a: To characterize the full anatomic and functional spectrum of complete AVSD.

SubAim 1b: Identify anatomic relationships and develop novel indices to improve the ability discriminate between unbalanced and balanced CAVSD.

Aim 2(A2): Determine patient and morphologic/physiologic factors that are associated with selection of surgical strategy (single ventricle repair, biventricular repair, intermediate (pulmonary artery banding)) and survival.

Aim3 (A3): Determine relationships between patient and anatomic characteristics, selected surgical strategy, and outcome.

Aim$ (A4): Develop and evaluate a clinically applicable prediction model to facilitate clinical decision making.

3. Background and Rationale

The AVSD is a spectrum of disease characterized by varying degrees of incomplete development of the septal tissue surrounding the atrioventricular valves along with varying degrees of abnormalities of the atrioventricular valves themselves. The essence of this cardiac malformation is a common atrioventricular junction [1]. The AVSD may be subcategorized into several groups of lesions (Appendix 1) [1, 2, 3, 4]. A complete AVSD is defined as an AVSD with a common AV valve and both a large defect in the atrial septum just above the AV valve (ostium primum ASD [a usually crescent-shaped ASD located between the antero-inferior margin of the fossa ovalis and the atrioventricular valves]) and a nonrestrictive defect in the ventricular septum just below the AV valve (in the canal (posterior) portion of the ventricular septum). The AV valve is one valve that bridges both the right and left sides of the heart.

Identification and surgical treatment of balanced AVSD is straightforward with excellent outcomes in the majority of cases [9, 10, 11, 12]. Unbalanced AVSD, however, encompasses a broad array of complex anatomies that present significant diagnostic and therapeutic challenges. These anatomic variations include right or left ventricular dominance, malalignment of the atrial and/or ventricular septa, variations in ventricular septal defect morphology, and abnormalities of the atrioventricular valve apparatus with potential abnormalities of atrioventricular valve function. The frequent incidence of comorbidities such as Trisomy 21 or heterotaxy syndrome, with its attendant pulmonary and systemic venous anomalies, further complicate the anatomic and therapeutic considerations coincident to uAVSD. Indeed, the very definition of what constitutes ‘unbalanced’ in AVSD is not well established. In addition, several surgical approaches are available to address the various anatomic substrates of uAVSD, including biventricular repair, single ventricle palliation, and “one and a half ventricle” repairs. Finally, there is scant literature documenting outcomes associated with these approaches. Historically, reported outcomes have been suboptimal with mortality approaching 25% in some series and low freedom from reintervention [8, 13, 14, 15]. Still fewer reports directly address surgical decision-making [16]. To achieve significant improvement in the treatment of uAVSD, clearer understanding of the morphologic and physiologic aspect of the disease and diagnostic criteria is needed [17]. Once diagnostic clarity is achieved, then the relationships between morphology/physiology and treatment strategies can be explored with the expectation of arriving at inferences that can guide clinical decisions going forward.

As a precursor to a proposed prospective study, a multi-institutional retrospective study was undertaken. This study had three principle aims: to measure the incidence of uAVSD, to determine early mortality rates associated with surgical repair of uAVSD, and to validate a reliable echocardiographic measurement that could be utilized as an enrollment tool for a prospective study. The atrioventricular valve index (AVVI) was originally introduced by Cohen and colleagues [18] as an echocardiographic measure of unbalance in AVSD, but it remains underutilized as a diagnostic criterion.

The original AVVI was calculated using the echocardiographic subcostal left anterior oblique (LAO view) to measure the area of common AVV apportioned over each ventricle and calculating the ratio of the smaller AV valve area over the larger AV valve area, so that left-dominant uAVSD was expressed as RAVV area/LAVV area and right dominant uAVSD the inverse [17, 18,].

The modified AVVI also is calculated using the echocardiographic subcostal left anterior oblique (LAO view) to measure the area of common AVV apportioned over each ventricle but is calculated by determining the ratio of the left atrioventricular valve area divided by the total atrioventricular valve area (LAVV area/ Total AVV area) [17, 20].

We modified the expression of AVVI (mAVVI) to simplify its use and evaluated its utility as an enrollment tool.

The major findings of the retrospective study are:

  1. Incidence of uAVSD amongst a cohort of complete AVSD who had a measurable AVVI and had surgery was 19% (58/305).
  1. Early mortality amongst the uAVSD cohort was 22.4% [UVR 7/22 (32%), BVR 4/34 (12%), PA Band 1/1 (100%)]. Notably, early mortality amongst the balanced AVSD cohort was 6.9% (17/247). UVR=Univentricular Repair. BVR=Biventricular Repair.
  1. Modified AVVI proved a reproducible and reliable method for identifying uAVSD from a cohort of all complete AVSD’s. All patients with mAVVI <0.2 underwent UVR, and nearly all patients with mAVVI 0.4 – 0.6 underwent BVR. Heterogeneity of surgical strategy was found among patients with mAVVI 0.19 – 0.39. There was a notable clustering of mortality within this range of mAVVI. This was true whether patients were undergoing UVR or BVR within that range of AVVI. For an AVVI between 0.19 and 0.39 (N=38), 26 patients underwent BVR (7 deaths), and 12 patients underwent UVR (4 deaths).

It is important to understandthat, for the retrospective study,the range of mAVVI chosen to define balanced AVSD was selected a priori by the investigators. That is, a mAVVI of 0.4 – 0.6 was called “balanced” (0.1 to either side of the middle – 0.5). This range of mAVVI was found to be reasonably concordant with outcome and surgical decision-making, as stated above. However, this construct was chosen as a starting point, and could conceivably change in light of new data obtained during the prospective study. In addition, there may be other echocardiographic measurements that are essential to proper assignation of “unbalance”, such as ventricular volumetric surrogates or septal malalignment assessments, and mAVVI should therefore be considered an importantcomponent of the anatomic landscape of AVSD rather than a measure that defines unbalance. Therefore, the exact range of mAVVI that corresponds to “balanced” AVSD remains to be established and also to be determined is whether otherfactors that, when present, impact the breadth of this range.

This prospective study aims to establish the echocardiographic indices, as well as patient anatomic and physiologic factors that favor BVR and those that favor UVR. These elements will be used to create a prediction model that facilitates optimal patient and procedure matching, thereby maximizing clinical outcomes. A hypothetical construct of the elements relevant to such a prediction model follows.

FAVORING BVR / FAVORING UVR
Favorable mAVVI / Unfavorable mAVVI
Favorable AV valve color inflow quantification / Unfavorable AV valve color inflow quantification
Favorable Ventricular volume measurement / Unfavorable ventricular volume measurement
Favorable septal alignment / Unfavorable septal alignment
Favorable leaflet geometry / Unfavorable leaflet geometry
Significant common AV valve regurgitation / Competent common AV valve
Poor ventricular function / Better ventricular function
Elevated PA pressures / Normal PA pressures
Echo measures of elevated EDP / Echo measures of normal EDP
Larger RV/LV inflow angle / Smaller RV/LV inflow angle

4.Study Design

4.a. Enrollment: All subjects diagnosed with complete AVSD at participating CHSS institutions will be considered for enrollment in the study. Patients who have undergone prior cardiac surgery at a non-CHSS institution will not be included. Informed consent, enrollment form completion, and data collection will be carried out by the participating center. A copy of the signed consent and data release form will be sent to The CHSS Data Center along with patient charts and imaging records. Data collection will be ongoing from enrollment forward, with submission of data annually and after any surgical intervention. Data submitted will include clinical records, echocardiograms, and procedural records. Data will be de-identified and securely stored at the CHSS Data Center at The Hospital for Sick Children, Toronto, Canada. The data will be abstracted by the Data Center staff. In addition to data submission from the participating center, annual phone follow-up will be performed by Data Center staff. The patients will be followed for their life. The study will be considered completed when the last enrolled patient is known to have died.

Patients who meet the eligibility criteria but are known to have passed away before being consented will be enrolled in the study. However their families will not be contacted for consent, follow-up or for any other study related purpose.

4.b. Echo Core Lab: All echocardiograms will be reviewed by a dedicated team of echocardiographers who will constitute the echo core lab. This will be a “virtual” core lab which is structured using a password protected encrypted software platform. Echocardiograms will be loaded into the web-based system. Use of a web-based system will enable expedited echocardiographic interpretation and review by multiple echocardiographers without the logistical issues and cost attendant to shipping of echos or travel for echo personnel. The echo core lab team will confirm the diagnosis, independently measure a mAVVI, and perform additional echo analyses as required for the study. A separate echocardiographic cohort of subjects undergoing three dimensional echo studies will also be accrued from participating centers performing such studies.

5. Participation Criteria

Participation in this cohort is voluntary for member institutions and surgeons. In an effort to encourage consistent enrollment of study patients by member institutions, foster active surgeon involvement, and ensure imaging studies of sufficient quality for analysis, each participating center will expected to support the following initiatives:

5.a. Echocardiography Training: Echocardiograms must be of sufficient quality and completeness that all study measurements may be performed by the echo core lab personnel. To ensure this quality and completeness, each participating center will send at least onelead sonographer and one echocardiographer (M.D.) to a one day educational seminar hosted by the echo core lab team. The number and location of these seminars is to be determined, but will most likely be held at the DataCenter or in conjunction with national pediatric cardiology meetings. These seminars will be recurring to facilitate accrual of additional participating sites, and their cost to the participating site covered under the study budget.

5.b. Member Participation: It is expected that a surgeon representative of the participating center will attend a minimum of one project related working session, either at a CHSS “work weekend” or at a national meeting where work on the study is being conducted.

6. Study Population

6.a. Inclusion Criteria

  1. Diagnosis of or referral with complete AVSD at a CHSS member institution within first year of life.

(Includes Tetralogy of Fallot or Double Outlet Right Ventricle with complete AVSD)

  1. Atrioventricular and Ventriculoarterial concordance (with the exception of DORV).
  2. Informed written consent.

6.b.Exclusion Criteria

  1. Partial or Transitional AVSD.
  2. Separate AV valve orifices
  3. Non-existent ventricular septal defect
  4. Aortic atresia
  5. First Intervention at a non-CHSS institution

6.c. Number of Subjects Projected: The retrospective study accrued approximately 350 patients over six years from four member institutions, roughly 60 of whom were considered to have uAVSD. Thus, each participating center may be expected to enroll an average of 10-15 patients per year. As such, between 300-450 subjects would be accrued within one year if one third to one half of CHSS members participated. We estimate a total of 1500 patients will be enrolled.

7. Study Endpoints and Evaluation

7.a. Primary Endpoints: The primary endpoints to be analyzed are surgical strategy, and early and late survival.

7.b. Secondary Endpoints: Secondary endpoints to be analyzed are:

  1. Echocardiographic measures of function including AV valve regurgitation or stenosis, ventricular function, and the presence or absence of left ventricular outflow tract obstruction.
  2. Unplanned reintervention after primary surgical repair
  3. Residual pulmonary hypertension (pulmonary artery pressure ≥ 2/3 systemic pressures).

7.c. Statistical Methods: Descriptive statistics will be calculated, including means with standard deviations, medians with ranges, 95% confidence intervals, and minimum, 25th percentile and 75th percentile (interquartile range), median, and maximum values for all continuous variables. Frequency counts and percentages will be used for categorical variables.

Cluster analysis using selected variables will be used to determine the appropriate number of differentiated groups of patients, followed by discriminant function analysis to determine which variables define the various groups of patients.

Competing risks methodology will be used to examine 4 end states until commitment: biventricular repair, univentricular repair, what we have termed “intermediate repairs”, and death prior to surgery. The bootstrap bagging method and risk analysis will be performed on each end state to determine associated factors. In addition, survival and hazard modeling will be performed from the time of commitment. A prediction nomogram will be developed based on the AVVI or otherechocardiographic measures (as determined by cluster and discriminant factor analysis), and utilized to predict given end states. These models will be used to assess different combinations of risk factors and determine the extent of the risk factors to help predict whether certain subject or management characteristics predict outcome.

Finally, surgical strategies will be analyzed and a prediction model generated which will predict the most appropriate surgical strategy for a given set of patient and anatomic factors.Initial procedure strategy indicated will assess the number of intended univentricular repair or biventricular repairs. Parametric risk-hazard analysis will be used to identify predictors of death for univentricular and for biventricular repair, which will allow prediction of the 5-year univentricular survival advantage for every infant. Survival will be scrutinized for children managed discordantly to univentricular survival advantage predictions.